Background

Pesco Telecom’s customers had a hard time finding the right part with text search. Product names and model numbers are long, confusing, and often written in different ways across brands and suppliers.

Additionally, many users search from the field on their phones, where labels are scratched, tiny, or in low light, so they can’t type exact terms. Moreover, the catalog has many look-alike items (e.g., connectors, splitters, patch cords) with small visual differences, and the same part can have multiple SKUs or legacy codes, which makes text search fail.

Furthermore, language mix (English/Arabic abbreviations), typos, and missing metadata reduce search accuracy, leading to slow searches, more back-and-forth with support, and lost sales—especially during urgent maintenance or peak demand.

Finally, the support team spends extra time confirming part compatibility, which increases costs and delays orders.

Proposed Solution

Nebulane developed a visual search feature that enables customers to identify parts using photos instead of text-based queries. By selecting the “Take/Upload Photo” option, users can submit an image, which the system analyzes from multiple angles. It then automatically crops the relevant part, generates a visual “fingerprint,” and compares it against the product catalog.

The feature returns the closest matches, complete with product names, images, pricing, availability, and key specifications. Each result includes a simple confidence indicator (e.g., “High match”) to guide the user. Customers can refine their search using filters such as part type, connector, length, or brand, or opt to upload another photo if needed.

The system is optimized for mobile use, including scenarios with low lighting or worn-out labels. Results are delivered in under a few seconds and the infrastructure scales automatically during periods of high demand.

New products are indexed automatically, and user feedback (e.g., thumbs up/down) is used to continuously improve the accuracy of future matches. Basic safety mechanisms are in place to detect and block inappropriate images.

Operational metrics—including accuracy, response times, and conversion rates—are monitored through detailed logs and dashboards, enabling the team to continuously enhance performance and user experience.

Metrics for Success

The success of the visual search feature was evident through several key indicators. Customers were able to identify the correct parts more quickly and with less effort. Engagement with suggested products increased, as reflected in higher click-through and add-to-cart rates, while support requests and product returns due to incorrect selections declined.

Usage of the visual search was notably higher on mobile devices, accompanied by more positive user feedback regarding the overall experience. The website maintained stability and responsiveness during periods of peak traffic, contributing to improvements in both overall sales and customer satisfaction.

Conclusion

The implementation of visual search significantly enhanced the customer experience by reducing friction in product discovery and increasing engagement, particularly on mobile. The system’s speed, scalability, and intelligent feedback loop contributed to measurable improvements in accuracy, customer satisfaction, and sales performance. These results validate visual search as a high-impact capability and a strong foundation for further innovation in product identification and e-commerce UX.